DOI: 10.1148/radiol.2421050677
(Radiology 2007;242:94-100.)
© RSNA, 2007
Role of Functional MR in Determining Language Dominance in Epilepsy and Nonepilepsy Populations: A Bayesian Analysis1
L. Santiago Medina, MD, MPH,
Byron Bernal, MD and
Jeniffer Ruiz, RN
1 From the Department of Radiology and Health Outcomes, Policy and Economics (HOPE) Center, Miami Children's Hospital, 3100 SW 62nd Ave, Miami, FL 33155. Received April 25, 2005; revision requested June 30; final revision received September 9; accepted October 5; final version accepted April 17, 2006.
Address correspondence to L.S.M. (e-mail: smedina{at}post.harvard.edu, santiago.medina{at}mch.com).
 |
ABSTRACT
|
|---|
Purpose: To determine the role of functional magnetic resonance (MR) imaging in assessing hemispheric language dominance in epilepsy and nonepilepsy populations.
Materials and Methods: A Bayesian analysis study was performed. The study was based on data from the published literature; thus, institutional review board approval was not required. The literature was reviewed to obtain pretest probabilities and likelihood ratios, which were analyzed to determine the posttest probability of language dominance by using functional MR imaging. Pretest probabilities of different hand dominances in epilepsy and nonepilepsy populations were obtained from the largest studies available in the literature. Likelihood ratios were derived from the sensitivity and specificity of functional MR imaging by using electrocortical stimulation (ECS) and the Wada test as reference standards.
Results: Likelihood ratios for functional MR in determining language dominance were 9.3 and 12.3 with ECS and the Wada test as reference standards, respectively. Use of functional MR increased the final posttest probabilities of hemispheric language dominance in epilepsy and nonepilepsy populations. In the nonepilepsy population, regardless of hand dominance, there was very high posttest probability (
96%) of a correlation between functional MR hemisphere activation and definite left-hemisphere language dominance. In the epilepsy population with right-hand dominance or ambidexterity, there was very high posttest probability (95%) of a correlation between functional MR hemisphere activation and definite left-hemisphere language dominance. In the left-handed nonepileptic subjects, there was high posttest probability (81%–83%) of a correlation between functional MR hemisphere activation and definite right-hemisphere language dominance. In the left-handed epilepsy population, there was high posttest probability (80%–97%) of a correlation between functional MR hemisphere activation and definite left-hemisphere language dominance. In the epilepsy population with ambidexterity, there was high posttest probability (80%–87%) of correlations between functional MR hemisphere activation and both definite right-hemisphere and bilateral language dominance.
Conclusion: Use of functional MR increases importantly the posttest probabilities of hemispheric language dominance in multiple subgroups of individuals with and without epilepsy.
© RSNA, 2007
|
INTRODUCTION
|
|---|
For more than a decade, functional magnetic resonance (MR) imaging has been used as a noninvasive test to determine language lateralization in subjects with and those without epilepsy (1). In a recent study by Medina et al (2), functional MR results influenced the diagnostic and therapeutic decisions of the seizure team: Language dominance changed, confidence in the identification of critical brain function areas increased, patient and family counseling was altered, and intraoperative mapping and surgical approaches were modified. At many institutions, however, the Wada test and electrocortical stimulation (ECS) continue to be the primary language lateralization tests because they have been in clinical use for a longer period and are still considered the reference standards (1). However, these tests have substantial shortcomings: The Wada test is especially limited when the right hemisphere is dominant. A study by Wyllie et al (3) revealed two false-positive right hemisphere–dominant results with use of the Wada test in nine patients with epilepsy (78% specificity). Because ECS rarely is performed in both hemispheres, partial or bilateral language dominance cannot be determined. Proponents of functional MR imaging have also advocated using this examination rather the Wada test or ECS because these latter tests are invasive and more expensive (4).
Investigators in multiple studies have compared the diagnostic performance of functional MR imaging with that of the Wada test or ECS (Table 1) (1,5–17). These studies have revealed high agreement between functional MR imaging findings and either Wada test or ECS results. However, there is no clear understanding of when functional MR imaging may be able to replace these more invasive tests. The role of functional MR imaging in language lateralization may be unclear because of the effect that two variables—hand dominance and epilepsy versus nonepilepsy subject population—have on hemispheric dominance.
View this table:
[in this window]
[in a new window]
|
Table 1. Published Studies of Concordance of Language Lateralization between Functional MR Imaging and either Wada Test or Direct ECS
|
|
We hypothesized that use of functional MR imaging increases importantly the posttest probabilities of hemispheric language lateralization in epileptic and nonepileptic individuals with different hand dominances. Thus, the purpose of our study was to determine the role of functional MR imaging in hemispheric language dominance in epilepsy and nonepilepsy populations.
|
MATERIALS AND METHODS
|
|---|
Study Design
Bayesian analysis was performed by using the well-established Bayes theorem (18,19). Physicians can learn about interpreting diagnostic tests by applying the Bayes theorem to a variety of clinical situations (19). Bayesian analysis is based on three key components: pretest probability, likelihood ratio, and posttest probability. The Bayes theorem formula to determine the posttest probability of hemispheric dominance, given a positive functional MR result (20,21), is as follows:
 |
where P indicates probability; T+, positive test result, D+, hemispheric language dominance; and D–, no hemispheric language dominance. This equation reflects the posttest probability once the pretest probability and likelihood are computed mathematically. Pretest probability is the percentage of patients or subjects with a specific language dominance in an epilepsy or nonepilepsy population. For example, 96% of right-handed nonepileptic subjects reportedly have left-hemisphere language dominance (22). The likelihood ratio, the information gleaned from a diagnostic test, is determined from the sensitivity and specificity of an examination. For example, the sensitivities and specificities of functional MR imaging range between 90.3% and 92.5%, with likelihood ratios of 9.3–12.3 (Table 2). Once the pretest probability and the likelihood ratio are available, the posttest probability can be determined by using the Bayes theorem. This Bayesian analysis approach is summarized in the Figure. Our hospital institutional review board did not require its approval for the Bayesian analysis studies, as they were based on data from the available literature.
View this table:
[in this window]
[in a new window]
|
Table 2. Sensitivity, Specificity, and Likelihood Ratios for Functional MR Language Lateralization, with Wada Test or ECS as Reference Standard
|
|
View larger version (22K):
[in this window]
[in a new window]
[Download PPT slide]
|
Bayesian analysis nomogram based on the Bayes theorem. Pretest and posttest probabilities are shown as they relate to the likelihood ratio. For example, the dotted line (A) shows a pretest probability of 70% for a specific language dominance increasing to greater than 95% after functional MR imaging; the likelihood ratio is 9.3.
|
|
Pretest Probability
The published literature was systematically reviewed to determine the pretest probabilities of language dominance in epilepsy and nonepilepsy populations (L.S.M., B.B., J.R.) (23). The authors used PubMed (National Library of Medicine, Bethesda, Md) to perform a Medline search for 1990–2004 data relevant to language dominance in epileptic and nonepileptic individuals. The main key words in the search strategy were language lateralization OR language dominance, combined with each of the following logic operands: AND Wada, AND cortical electrical stimulations, AND NIS (Near Infrared Spectroscopy), AND transcranial magneto-stimulation, and AND transcranial Doppler. Animal studies and non–English-language articles were excluded in this search strategy. The references in the articles were reviewed to complement the literature search.
The search strategy revealed 54 articles for review. The criteria for inclusion as one of the systematically reviewed pretest probability studies were (a) clear information about the reference standard used for all subjects or patients, (b) clear definition of the side of language dominance in all subjects or patients, and (c) inclusion of 50 or more subjects or patients. All articles that fulfilled these criteria are shown in Table 3. The pretest probability of language dominance in the nonepilepsy population was based on the results of noninvasive procedures performed in healthy subjects (22,24). There were clear differences in language dominance based on the two important variables: (a) hand dominance and (b) epilepsy versus nonepilepsy population (Table 3). Each study was stratified to allow analysis of each published article (Table 3).
Likelihood Ratio
The published literature was systematically reviewed to determine the agreement between functional MR findings and Wada test or ECS results in language lateralization (L.S.M., B.B., J.R.) (Table 1) (23). The authors used PubMed to perform a Medline search for 1994–2004 data. The main key words in the search strategy were functional MR imaging OR functional MRI, combined with each of the following logic operands: AND (Wada OR amobarbital OR amytal), AND electrical cortical stimulation, and AND DCES (direct electrical cortical stimulations). Animal studies and non–English-language articles were excluded in this search strategy. The references in the articles were reviewed to complement the literature search.
The search strategy revealed 57 articles for review. The criteria for inclusion as one of the systematically reviewed likelihood ratio studies were (a) information about true-positive, true-negative, false-positive, and false-negative results; (b) similar functional MR language tasks and paradigms; and (c) inclusion of five or more patients. All articles that fulfilled these criteria are shown in Table 1. Thirteen studies, with a total of 240 patients with functional MR–Wada test correlation and a total of 31 patients with functional MR–ECS correlation, were identified (Table 1). With data from these pooled studies, a 2 x 2 contingency table was generated by determining the true-positive, true-negative, false-positive, and false-negative results on the basis of agreement and disagreement between functional MR and the reference standard—either Wada testing or ECS—regarding hemispheric language dominance. Although not perfect, the Wada test and ECS are the best invasive reference standards available for determining language lateralization. The sensitivity and specificity of functional MR imaging in language lateralization were determined mathematically (Table 2) (27). Ninety-five percent CIs were calculated for sensitivity, specificity, and likelihood ratio (28,29).
Posttest Probability
By using the Bayesian analysis nomogram (Figure), we determined posttest probabilities according to hand dominance and patient population (epilepsy vs nonepilepsy) (L.S.M., J.R.). Numbers were rounded to the nearest percentage.
|
RESULTS
|
|---|
With pooled Wada test data as the reference standard, both the sensitivity and the specificity of functional MR imaging were 92.5% (95% CI: 89.1%, 95.9%) and the likelihood ratio was 12.3. With pooled ECS data as the reference standard, both the sensitivity and the specificity of functional MR imaging were 90.3% (95% CI: 80%, 100%) and the likelihood ratio was 9.3. Use of functional MR imaging increased the posttest probability for all hand dominance groups in both the epilepsy and the nonepilepsy populations (Tables 4 and 5).
Nonepilepsy Population
In the nonepilepsy population, regardless of hand dominance, if functional MR imaging showed left hemisphere–dominant function, the posttest probability was 96% or greater that the left hemisphere was in fact dominant. In the nonepilepsy population with right-hand dominance, if functional MR imaging showed left hemisphere–dominant function, the posttest probability was greater than or equal to 99% that the left hemisphere was dominant. However, if functional MR imaging showed right hemisphere–dominant function, the posttest probability was less than 65% that the right hemisphere was dominant. In the nonepilepsy population with left-hand dominance or ambidexterity, if functional MR imaging showed right hemisphere–dominant function, the pretest probabilities of 26%–27% and 15%–57% increased accordingly to posttest probabilities of 81%–83% and 60%–95%, respectively, that the right hemisphere was dominant.
Epilepsy Population
In the epilepsy population, hand dominance and functional MR findings were important in determining posttest probabilities of language lateralization in the different subgroups. In epileptic patients with right-hand dominance or ambidexterity, if functional MR imaging showed left hemisphere–dominant function, the posttest probability was 95% or greater that the left hemisphere was in fact dominant. In left-handed epileptic patients, functional MR left-dominant activation increased the 33%–75% pretest probability of left-hemisphere dominance to a posttest probability of 80%–97% and functional MR right-dominant activation increased the 18%–40% pretest probability of right-hemisphere dominance to a posttest probability of 68%–90%. In the ambidextrous epilepsy population, both functional MR right-dominant activation and functional MR bilateral activation increased the 33% pretest probability to a posttest probability of 80%–87% (Table 4).
|
DISCUSSION
|
|---|
Our Bayesian analysis revealed that use of functional MR imaging increased the posttest probabilities of language lateralization in both epilepsy and nonepilepsy populations. The magnitude of the effect of functional MR imaging on the degree of posttest probability depended on two important variables: the population studied (epilepsy vs nonepilepsy) and hand dominance. By factoring these two variables and the diagnostic information gleaned from functional MR imaging, we identified two subgroups: individuals with very high (>95%) and those with high (80%–97%) probability of a correlation between side of functional MR activation and definite language hemisphere dominance.
In the nonepilepsy population, regardless of hand dominance, there was very high posttest probability (96%) of a correlation between functional MR hemisphere activation and definite left-hemisphere language dominance. In the epilepsy population with right-hand dominance or ambidexterity, there was very high posttest probability (95%) of a correlation between functional MR hemisphere activation and definite left-hemisphere language dominance. In the nonepilepsy population with left-hand dominance, there was high posttest probability (81%–83%) of a correlation between functional MR hemisphere activation and definite right-hemisphere language dominance. In the epilepsy population with left-hand dominance, there was high posttest probability (80%–97%) of a correlation between functional MR hemisphere activation and definite left-hemisphere language dominance. In the epilepsy population with ambidexterity, there was high posttest probability of correlations between functional MR hemisphere activation and both definite right (80%–87%) and bilateral (80%–87%) language dominance.
In contrast to the very-high-probability and high-probability groups just described was the group of subjects with low (<65%) posttest probabilities—specifically, the right-handed nonepileptic subjects with right-hemisphere language dominance. Right-hemisphere functional MR activation increased the posttest probability to only 24%–64%. For this subgroup, another examination such as the Wada test or ECS is clearly required to determine language dominance.
All available tests to determine language dominance have pros and cons. Functional MR imaging is a noninvasive examination that can be used not only to determine hemispheric language dominance but also for anatomic mapping of areas with important function (1). Yetkin et al (30) observed 100% correlations between the intraoperative site and the functional MR activation site within 20 mm and 87% correlations within 10 mm. However, functional MR imaging has limitations when there is susceptibility artifact (ie, ferromagnetic material from recent surgery or hematoma) and when the patient has had seizures recently (27). The Wada test is an invasive test that only reveals hemispheric language dominance, without precise anatomic mapping of function (1,4). ECS, another invasive test, enables detailed superficial anatomic mapping of brain function (1). However, it is limited in the evaluation of function in deep areas of the brain and rarely is performed in both hemispheres; thus, partial or bilateral language dominance cannot be determined (3).
Our Bayesian analysis had limitations. The pretest probabilities of language dominance in the nonepilepsy population were based on the results of noninvasive procedures performed in healthy volunteers (22,24). The pretest probabilities of language dominance in the epilepsy population were based on the results of invasive procedures: ECS and bilateral Wada testing (3,20,21,25). Although the pretest probabilities determined in these studies have limitations, we selected those studies with the best study design and the largest populations to obtain the most robust data available (3,21,22,24–26).
There were limitations in determining the sensitivities, specificities, and likelihood ratios for functional MR imaging. However, we presented ranges of sensitivity, specificity, and likelihood ratio values by using the two best available reference standards: ECS and the Wada test. The true diagnostic performance and likelihood ratio of functional MR imaging probably lie between those of these two reference-standard tests. Therefore, all results are presented as ranges to give the reader a clear understanding of the variability.
In summary, use of noninvasive functional MR imaging increases importantly the posttest probabilities for language hemispheric dominance in multiple subgroups in epilepsy and nonepilepsy populations. Rational use of pretest probabilities, likelihood ratios, and posttest probabilities for different populations allows a more complete understanding of the role of functional MR imaging in determining hemispheric language dominance.
|
ADVANCE IN KNOWLEDGE
|
|---|
- Noninvasive functional MR imaging, by increasing the posttest probabilities for language hemispheric dominance in multiple subgroups in epilepsy and nonepilepsy populations, has an important clinical role.
|
ACKNOWLEDGMENTS
|
|---|
We acknowledge Nolan R. Altman, MD, for his valuable comments regarding this article.
|
FOOTNOTES
|
|---|
Abbreviations: CI = confidence interval • ECS = electrocortical stimulation
Authors stated no financial relationship to disclose.
Author contributions: Guarantor of integrity of entire study, L.S.M.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript final version approval, all authors; literature research, all authors; clinical studies, B.B.; statistical analysis, L.S.M., J.R.; and manuscript editing, all authors
|
References
|
|---|
- Lurito J, Bryan R, Vincent P, Ulmer JL. 2000 Syllabus: categorical course in diagnostic radiology—neuroradiology: functional brain mapping. Oak Brook, Ill: Radiological Society of North America, 2000; 79–104.
- Medina LS, Bernal B, Dunoyer C, et al. Seizure disorders: functional MR imaging for diagnostic evaluation and surgical treatment—prospective study. Radiology 2005;236:247–253.[Abstract/Free Full Text]
- Wyllie E, Luders H, Murphy D. Intracarotid amobarbital (Wada) test for language dominance: correlation with results of cortical stimulation. Epilepsia 1990;31(2):156–161.[Medline]
- Medina LS, Aguirre E, Bernal B, Altman NR. Functional MR imaging versus Wada test for evaluation of language lateralization: cost analysis. Radiology 2004;230:49–54.[Abstract/Free Full Text]
- Desmond JE, Sum JM, Wagner AD, et al. Functional MRI measurement of language lateralization in Wada-tested patients. Brain 1995;118:1411–1419.[Abstract/Free Full Text]
- Binder JR, Swanson SJ, Wagner AD, et al. Determination of language dominance using functional MRI: a comparison with Wada test. Neurology 1996;46:978–984.[Abstract/Free Full Text]
- Hertz-Pannier L, Gaillar WD, Mott SH, et al. Noninvasive assessment of language dominance in children and adolescents with functional MRI: a preliminary study. Neurology 1997;48:1003–1012.[Abstract]
- Worthington C, Vincent DJ, Bryant AE, et al. Comparison of functional magnetic resonance imaging or language localization and intracarotid speech amytal testing in presurgical evaluation for intractable epilepsy. Stereotact Funct Neurosurg 1997;69:197–201.[Medline]
- Bahn MM, Lin W, Silbergeld DL. Localization of language cortices by functional MR imaging compared with intracarotid amobarbital hemispheric sedation. AJR Am J Roentgenol 1997;169:575–579.[Abstract/Free Full Text]
- Schlosser MJ, Luby M, Spencer DD, et al. Comparative localization of auditory comprehension by using functional magnetic resonance imaging and cortical stimulation. J Neurosurg 1999;91:626–635.[Medline]
- Yetkin FZ, Swanson S, Fischer M. Functional MR of frontal lobe activation: comparison with Wada language results. AJNR Am J Neuroradiol 1998;19:1095–1098.[Abstract]
- Benson RR, Fitzgerald DB, LeSueur LL. Language dominance determined by whole brain functional MRI in patients with brain lesions. Neurology 1999;52:798–809.[Abstract/Free Full Text]
- Hirsch J, Ruge MI, Kim KH. An integrated functional magnetic resonance imaging procedure for preoperative mapping of cortical areas associated with brain lesions. Neurosurgery 2000;47:711–721.[CrossRef][Medline]
- Woermann FG, Jokeit H, Luerding R, et al. Language lateralization by Wada test and fMRI in 100 patients with epilepsy. Neurology 2003;61:699–701.[Abstract/Free Full Text]
- Baciu M, Kahane P, Minotti L, et al. Functional MRI assessment of the hemispheric predominance for language in epileptic patients using a simple rhyme detection task. Epileptic Disord 2001;3:117–124.[Medline]
- Gaillard WD. Language dominance in partial epilepsy patients identified with an fMRI reading task. Neurology 2002;59:256–265.[Abstract/Free Full Text]
- Rutten GJ. fMRI-determined language lateralization in patients with unilateral or mixed language dominance according to the Wada test. Neuroimage 2002;17:447–460.[CrossRef][Medline]
- Weinstein MC, Fineberg HV. Clinical decision analysis. Toronto, Ontario, Canada: Saunders, 1980; 75–130.
- Sox HC Jr, Blatt MA, Higgins MC, Marton KI. Medical decision making. Boston, Mass: Butterworth-Heinmann, 1988; 67–101.
- Jayakar P, Bernal B, Medina SL, Altman NA. False lateralization of language cortex on functional MRI after a cluster of focal seizures. Neurology 2002;58:490–492.[Abstract/Free Full Text]
- Kurthen M, Helmstaedter C, Linke D. Quantitative and qualitative evaluation of patterns of cerebral language dominance. Brain Lang 1994;46:536–564.[CrossRef][Medline]
- Knecht S, Drager B, Deppe M, et al. Handedness and hemispheric language dominance in healthy humans. Brain 2000;123(pt 12):2512–2518.[Abstract/Free Full Text]
- Deeks JJ. Systematic reviews in health care: systematic reviews of evaluations of diagnostic and screening tests. BMJ 2001;323(7305):157–162.[Free Full Text]
- Khedr EM, Hamed E, Said A, Basahi J. Handedness and language cerebral lateralization. Eur J Appl Physiol 2002;87:469–473.[CrossRef][Medline]
- Risse GL, Gates JR, Fangman MC. A reconsideration of bilateral language representation based on the intracarotid amobarbital procedure. Brain Cogn 1997;33:118–132.[CrossRef][Medline]
- Loring DW, Meador KJ, Lee GP, et al. Cerebral language lateralization: evidence from intracarotid amobarbital testing. Neuropsychologia 1990;28:831–838.[CrossRef][Medline]
- Medina LS. Study design and analysis in neuroradiology: a practical approach. AJNR Am J Neuroradiol 1999;20:1584–1596.[Free Full Text]
- Medina LS, Zurakowski D. Measurement variability and confidence intervals in medicine: why should radiologists care? Radiology 2003;226:297–301.[Abstract/Free Full Text]
- Bland M. An introduction to medical statistics. Oxford, England: Oxford Medical Publications, 1987.
- Yetkin FZ, Wade M, Mueller M. Functional MR activation correlated with intraoperative cortical mapping. AJNR Am J Neuroradiol 1997;18:1311–1315.[Abstract]
TOP
ABSTRACT
INTRODUCTION
